| { | |
| "paper_id": "1981", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T07:47:12.510176Z" | |
| }, | |
| "title": "PRACTICAL EXPERIENCE OF MACHINE TRANSLATION V. Lawson (ed.) 53 North-Holland Publishing Company / \u00a9 ASLIB, 1982 DISCUSSION 0N SESSION 1: Translation in Transition", | |
| "authors": [ | |
| { | |
| "first": "Nigel", | |
| "middle": [], | |
| "last": "Bevan", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "National Physical Laboratory", | |
| "institution": "", | |
| "location": {} | |
| }, | |
| "email": "" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "", | |
| "pdf_parse": { | |
| "paper_id": "1981", | |
| "_pdf_hash": "", | |
| "abstract": [], | |
| "body_text": [ | |
| { | |
| "text": "It was suggested that MT could encourage sloppy translations. Translators would often accept poor translations produced by the machine rather than correct them. It was agreed that this may be true, although it is much easier to have second thoughts when using a word processor, than it is with the final draft of a typed copy.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "1.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "A human translator varies his style, while MT always uses the same words, which can produce pedestrian and boring translations. The predictability of MT was considered an advantage for technical translations, where consistent terminology is a cardinal principle.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "2.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "20% of METEO translations are rejected and require human translation. These are due to input errors and unacceptable sentence structures, and no further improvement is possible without fundamentally altering the system.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "3.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "On the Mitel system 85% of translators' time in the first 2 -3 months was spent on dictionary updates, although it is now only 5% (3 -4 terms a day). Size of dictionary is not necessarily a good guide to accuracy of translation, as it depends on the degree of specialisation of the user. The METEO dictionary requires only 1600 entries.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "4.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "The Mitel system has reduced costs, and produces translations acceptable to the end user, although there are no precise figures.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "5.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "The Federal Translation Bureau in Canada recently compared the costs of several MT systems, and found Weidner slightly cheaper than SYSTRAN for the specific types of text evaluated. A new evaluation is under way using the latest versions of the systems. METEO, with its limited vocabulary and sentence structures, is considerable cheaper.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "b.", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "All systems suffer if the input contains spelling errors or badly constructed sentences. Closer control of the input was considered important, and it was pointed out that many word processing systems contain dictionaries which can be used to highlight probable spelling errors.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "7.", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": {}, | |
| "ref_entries": {} | |
| } | |
| } |